Quantitative spectrophotometry of scattering media via frequency-domain and constant-intensity measurements

Use of HSI to measure oxygen saturation in the lower limb and its correlation with healing of foot ulcers in diabetes

AIM

To explore the use of hyperspectral imaging (HSI) to predict healing of diabetic foot ulcers in patients with diabetes.

METHODS

We used a HSI technique that incorporated novel software to account for tissue scattering of light, and was validated using blood samples of varying oxygen saturation assessed by blood gas analysis. HSI was then performed on a population newly presenting with diabetic foot ulcers to a specialist clinic, and associations were sought with healing at 12 and 24 weeks.

RESULTS

The correlation between the results of HSI and blood gas analysis was strong (r = 0.994). A total of 43 patients (mean ± sd age 62.7 ± 12.2 years; 31 men, 12 women; 37 with Type 2 diabetes, six with Type 1 diabetes) with foot ulcers were included in the prospective clinical study and underwent HSI within 16 days of presentation. In all, 26 ulcers healed within 12 weeks and 28 within 24 weeks. There was a negative association between tissue oxygenation assessed by HSI at baseline and healing by 12 weeks (P = 0.009), and this was observed in both infected and non-infected ulcers. There was a significant positive correlation between oxygenation assessed by HSI and time to healing (P = 0.03). No correlations were observed at 24 weeks.

CONCLUSIONS

These findings suggest that HSI may predict healing in routine practice. The fact that the correlation between HSI and healing was negative could be explained by HSI being a measure of oxygenation of haemoglobin and there may be an inverse relationship between this and the oxygenation of extravascular tissue in people with neuropathy and/or microvascular disease.

Determination of the validity of spectrophotometric measurements based upon cumulants of the temporal point-spread function

Spectrophotometric measurements of scattering media are nontrivial owing to the nonlinear relationship between the observed attenuation A and the level of absorption coefficient micro(a). Presented is a method, utilizing frequency-domain measurements, capable of determining the validity of a measured change in micro(a) based purely on the characteristics of a baseline temporal point-spread function. The order of cumulant used is shown to provide variation of the sensitivity of the method, and its use is demonstrated through Monte Carlo simulations of transmission measurements taken from a scattering slab.

Reduction of error in spectrophotometry of scattering media using polarization techniques

Scattering can result in erroneous determination of the concentrations of constituent absorbers in spectrophotometry. This is due to the relationship between attenuation and absorption coefficient becoming nonlinear; hence, the use of the Lambert-Beer law becomes invalid. It has previously been shown that application of polarization techniques can reduce these effects, resulting in a more linear relationship. Here we quantify the impact of this improvement on measurement of the ratio of concentrations for two general absorbing species and show that measurement using polarization-maintaining light is more accurate. This is performed using a generalized version of theory previously dependent on selection of isosbestic wavelengths. For the absorbing species and geometries considered here, the mean error on the estimation of absorber concentration ratio is 18.2% for the case of detection without polarization discrimination. When polarization-maintaining light is extracted, mean errors of 1.2% and 5.1% are achieved for linear and circular polarizations, respectively. The improvement provided by the polarization techniques is observed regardless of the illuminating wavelengths but is achieved at the expense of a reduced signal-to-noise ratio. Taking this into account, for the detection scheme considered with a detector well capacity of 4 x 10(5) electrons the improvement provided by linear polarization-maintaining light is reduced to a factor of 3.6 and for circular polarizations a factor of 2.2.